Receiving loop antenna design
 

I have been working on the BPL Interference issue.

One of my projects has been exploring ways by which ordinary (well,
competent anyway,) amateurs can make reasonably reliable measurements of
noise / interference using existing amateur station equipment or
equipment that is easy for amateurs to construct.

This has led me to search for a portable antenna of reasonably
predictable gain that can be used with a known HF SSB receiver.

I have had a hack at predicting the gain and antenna factor of a small
square untuned loop driving a 50 ohm load. The model is in a Mathcad
worksheet, but I have copied it to a gif file which you can view on my
website at http://www.vk1od.net/bpl/loop.mcd.gif . (The worksheet is
entirely in metric units.)

I would appreciate any comments / review on the model and calcs.

Owen

Owen wrote:

I hate replying to my own posts... but...

I should have reminded you that if you are having trouble viewing the
fig file because it has been zoomed to fit in the browser window, most
modern browsers allow you to zoom it up to 100% size.

In Windows Exploder, hold your cursor over the image until a little
"Expand" control appears, click the "Expand" control and there you go.

In Firefox, just click on the image.

Owen

Owen wrote:

I hate replying to my own posts... but...

I should have reminded you that if you are having trouble viewing the
gif file because it has been zoomed to fit in the browser window, most
modern browsers allow you to zoom it up to 100% size.

In Windows Exploder, hold your cursor over the image until a little
"Expand" control appears, click the "Expand" control and there you go.

In Firefox, just click on the image.

Owen

Doesn't the receiver input impedance come into it somewhere? Where is
it?
----
Reg.

===================================

"Owen" wrote in message
...

I have been working on the BPL Interference issue.

One of my projects has been exploring ways by which ordinary (well,
competent anyway,) amateurs can make reasonably reliable
measurements of
noise / interference using existing amateur station equipment or
equipment that is easy for amateurs to construct.

This has led me to search for a portable antenna of reasonably
predictable gain that can be used with a known HF SSB receiver.

I have had a hack at predicting the gain and antenna factor of a
small
square untuned loop driving a 50 ohm load. The model is in a Mathcad
worksheet, but I have copied it to a gif file which you can view on
my
website at http://www.vk1od.net/bpl/loop.mcd.gif . (The worksheet is
entirely in metric units.)

I would appreciate any comments / review on the model and calcs.

Owen

Doesn't the receiver input impedance come into it somewhere? Where
is
it?
----
Reg.

===================================

Sorry! Looked through it again and found load impedance = 50 ohms.
----
Reg.

"Owen" wrote in message
...

I have been working on the BPL Interference issue.

One of my projects has been exploring ways by which ordinary (well,
competent anyway,) amateurs can make reasonably reliable measurements of
noise / interference using existing amateur station equipment or equipment
that is easy for amateurs to construct.

This has led me to search for a portable antenna of reasonably predictable
gain that can be used with a known HF SSB receiver.

I have had a hack at predicting the gain and antenna factor of a small
square untuned loop driving a 50 ohm load. The model is in a Mathcad
worksheet, but I have copied it to a gif file which you can view on my
website at http://www.vk1od.net/bpl/loop.mcd.gif . (The worksheet is
entirely in metric units.)

I would appreciate any comments / review on the model and calcs.

Owen


Looks pretty decent, until the very end. Antenna Factor (AF) is the ratio of
the field strength voltage to the output VOLTAGE, not power, although you
did get the numbers right.

So practically, since the average ham has a receiver with a sensitivity in
the order of a microvolt, then your antenna limits your minimum discernable
signal level to around 65 uV/m. Maybe 100 uV/m to be on the safe side.
That's likely quite adequate for detecting BPL noise, but the real problem
is having the average ham get an anywhere near reasonably accurate
measurement of 100 uV. Your S meter just isn't good enough, so now you're
moving beyond the "average" ham's capability. Accurizing your receiver into
an RF microvoltmeter is a tough task, so maybe the best route is to use a
signal generator as a comparison standard. Old boatanchor signal generators
in the 7 MHz region are reasonably available, and their attenuators are a
lot better than their frequency stability and portability. g

I applaud your goals, but getting data accurate enough to toss into an
intelligent argument about BPL is a tough task. Good luck.

--
Ed
WB6WSN
El Cajon, CA USA

Ed Price wrote:

Looks pretty decent, until the very end. Antenna Factor (AF) is the ratio of
the field strength voltage to the output VOLTAGE, not power, although you
did get the numbers right.

I have looked at it and I can't see that I said "power" in relation to
Antenna Factor. Perhaps I am blind. (You didn't confuse the units dB/m
(dB per meter) with dBm (db wrt 1mW) did you?)

So practically, since the average ham has a receiver with a sensitivity in
the order of a microvolt, then your antenna limits your minimum discernable
signal level to around 65 uV/m. Maybe 100 uV/m to be on the safe side.

In fact, the technique calls for measuring signals on the rx from the
noise floor to about 20dB above it. The noise floor for receivers today
is typically -135dBm.

That's likely quite adequate for detecting BPL noise, but the real problem
is having the average ham get an anywhere near reasonably accurate
measurement of 100 uV. Your S meter just isn't good enough, so now you're
moving beyond the "average" ham's capability. Accurizing your receiver into
an RF microvoltmeter is a tough task, so maybe the best route is to use a
signal generator as a comparison standard. Old boatanchor signal generators
in the 7 MHz region are reasonably available, and their attenuators are a
lot better than their frequency stability and portability. g

No, the technique does not use an S-meter. In a nutshell, it uses Ed
Hare's (W1RFI) technique for calibrating the noise floor of the
receiver, using an external attenuator to keep the rx input below the
AGC threshold, and measuring the audio output with signal and the audio
output from rx internal noise as inputs to a calculation of the input
signal power. Applying external attenuator losses, feedline losses and
antenna factor allows calculation of field strength.

I applaud your goals, but getting data accurate enough to toss into an
intelligent argument about BPL is a tough task. Good luck.


I have gotten sidetracked here, my real interest is the completeness /
accuracy of the loop model.

Owen

A well thought out approach, nice job!

"Owen" wrote in message
...

I have gotten sidetracked here, my real interest is the completeness /
accuracy of the loop model.

Owen

If you find yourself short of sensitivity, try a tuned loop in the
style of a magloop and match the antenna to the receiver.

But whatever you adopt, accuracy will be limited by the uncertainty in
the amateur's receiver input impedance. This will change from band to
band and its actual value will be a matter of guesswork.

A receiver's input impedance can be masked with an attenuator. But
this further reduces sensitivity.

With amateur grade equipment, facilities and environment, expect a
measuring uncertainty in the region of 4 to 7 dB at 7 MHz. Which is
good enough for most amateur purposes and makes your precision
calculations, including conductor diameter and conductivity, not worth
the trouble.

All you need for calculation is enclosed loop area, loop inductance,
receiver impedance and a pocket calculator.

The uncertainty of a measurement is just as important as the value
itself. The only way to assess uncertainty is to compare with
professional-grade equipment. In which case, if professional grade is
obtainable, you can dump the amateur stuff.

I do like the way your calculations appeared on my screen with one
mouse click.
How do you do it?

----
Reg.

==================================

"Owen" wrote in message
...

I have been working on the BPL Interference issue.

One of my projects has been exploring ways by which ordinary (well,
competent anyway,) amateurs can make reasonably reliable
measurements of
noise / interference using existing amateur station equipment or
equipment that is easy for amateurs to construct.

This has led me to search for a portable antenna of reasonably
predictable gain that can be used with a known HF SSB receiver.

I have had a hack at predicting the gain and antenna factor of a
small
square untuned loop driving a 50 ohm load. The model is in a Mathcad
worksheet, but I have copied it to a gif file which you can view on
my
website at http://www.vk1od.net/bpl/loop.mcd.gif . (The worksheet is
entirely in metric units.)

I would appreciate any comments / review on the model and calcs.

Owen

Looks pretty decent, until the very end. Antenna Factor (AF) is the ratio
of the field strength voltage to the output VOLTAGE, not power, although
you did get the numbers right.

So practically, since the average ham has a receiver with a sensitivity in
the order of a microvolt, then your antenna limits your minimum
discernable signal level to around 65 uV/m. Maybe 100 uV/m to be on the
safe side. That's likely quite adequate for detecting BPL noise, but the
real problem is having the average ham get an anywhere near reasonably
accurate measurement of 100 uV. Your S meter just isn't good enough, so
now you're moving beyond the "average" ham's capability. Accurizing your
receiver into an RF microvoltmeter is a tough task, so maybe the best
route is to use a signal generator as a comparison standard. Old
boatanchor signal generators in the 7 MHz region are reasonably available,
and their attenuators are a lot better than their frequency stability and
portability. g

I applaud your goals, but getting data accurate enough to toss into an
intelligent argument about BPL is a tough task. Good luck.

--
Ed
WB6WSN
El Cajon, CA USA

In the BPL report at
http://www.ofcom.org.uk/research/tec...line/ascom.pdf
I noticed the system noise floor at about 10 dBuV/m (in 9 kHz). For the
tests they used an active bi-conical antenna. (By my calculations 10 dBuV/m
is about 9 uV(2.5 kHz BW) from a 40 m dipole at 7 MHz.) In the previously
mentioned report most of the BPL signals -- even at 1 meter from the
source -- is 60 dBuV/m. It seems your system with the loop will be much
less sensitive at about 100 uV/m (+40 dBuV/m).

Incidentally, when I attempted to save your web page of math, it was saved
as an ".mcd" document. Obviously I was not able to open it with Mathcad,
but will have to type it in by hand.

Might be interesting to replicate your results with NEC2.

Frank

Reg Edwards wrote:
If you find yourself short of sensitivity, try a tuned loop in the
style of a magloop and match the antenna to the receiver.


In concert with an SSB rx of noise floor -135dBm in 2KHz Effective noise
bandwidth (they are realistic figures for an IC706IIG for example), the
rx noise in 2KHz BW is 0.04uV. With an antenna with AF=36dB, the rx
noise floor translates to field strength of 2.5uV/m or 8dBuV/m.

Clearly, the "instrument" is not going to be suitable for measurements
below 11dBuV/m. However, measurements of the BPL systems on "trial" here
(Mitusbishi based on DS/2 45Mbps chipsets) showed field strengths of 45
to 65dBuV/m in 2Hz and a rx with this loop needs 20+dB of RF attenuation
to keep the interference below the AGC threshold (with the benefit of
stabilising the rx input Z somewhat).

But whatever you adopt, accuracy will be limited by the uncertainty in
the amateur's receiver input impedance. This will change from band to
band and its actual value will be a matter of guesswork.

Yes, I have been measuring the rx noise floor with 20dB of attenuation
to simulate the common measurement configuration.


A receiver's input impedance can be masked with an attenuator. But
this further reduces sensitivity.


As discussed.

With amateur grade equipment, facilities and environment, expect a
measuring uncertainty in the region of 4 to 7 dB at 7 MHz. Which is
good enough for most amateur purposes and makes your precision
calculations, including conductor diameter and conductivity, not worth
the trouble.

Given that the interference is 70dB above the ambient noise floor, we
don't need 1/10dB accuracy to demonstrate to regulators that there is a
problem


All you need for calculation is enclosed loop area, loop inductance,
receiver impedance and a pocket calculator.

The uncertainty of a measurement is just as important as the value
itself. The only way to assess uncertainty is to compare with
professional-grade equipment. In which case, if professional grade is
obtainable, you can dump the amateur stuff.

Understood. My view is that if professional grade EMC measurement kit is
available, we can use it to do a lower grade calibration of the amateur kit.


I do like the way your calculations appeared on my screen with one
mouse click.
How do you do it?

The model is in Mathcad, I copied it to the clipboard and pasted it into
Frontpage (my web editor) which finds only a useful format in the
clipboard and saves it as a GIF file (ie a graphics image).

Thanks for checking the model Reg, I think you are telling me it is more
precise than needs to be, but you haven't faulted it for accuracy.

Owen

Frank wrote:

In the BPL report at
http://www.ofcom.org.uk/research/tec...line/ascom.pdf
I noticed the system noise floor at about 10 dBuV/m (in 9 kHz). For the
tests they used an active bi-conical antenna. (By my calculations 10 dBuV/m
is about 9 uV(2.5 kHz BW) from a 40 m dipole at 7 MHz.) In the previously

Yes, Fig 8 shows about 10dBuV/m in 9KHz which interpolates to 5dBuV/m in
3KHz, and their measurements used a peak detector. On white noise, the
QP value would probably be 2 to 3dB lower.

I have made a large number of measurements at my home QTH (in a
residential neighbourhoos) using a half wave dipole and assuming an
average gain of -1.2dBi or an AF of -11.6dB/m and I regularly get
ambient noise readings down to around 0 to 3dBuV/m QP in 3KHz or
extrapolated to 9KHz BW, 5 to 8dBuV/m QP. Ambient noise is probably
lower than indicated by ITU P372-8!

mentioned report most of the BPL signals -- even at 1 meter from the
source -- is 60 dBuV/m. It seems your system with the loop will be much
less sensitive at about 100 uV/m (+40 dBuV/m).


See my response to Reg re the noise floor for the setup, I make it
around 8dbuV/m or 2.5uV/m. I don't pretend it can measure ambient noise,
but it can and has measured BPL interference at 40dBuV/m to 70dBuV/m.

Incidentally, when I attempted to save your web page of math, it was saved
as an ".mcd" document. Obviously I was not able to open it with Mathcad,
but will have to type it in by hand.

No you won't, I have posted a later version of the mathcad file to
http://www.vk1od.net/bpl/loop02.mcd .

The file you downloaded is an image (.gif) called loop.mcd.gif, and it
looks like your download process dropped the extension, or you hide the
extension on your machine. (Some software thinks that the first dot
begins the file extension, whereas it is the last dot that does so.)


Might be interesting to replicate your results with NEC2.

I have modeled the loop in EZNEC and get very similar inductance and
resistances.

Thanks Fred, appreciate the review.

Owen

One of the most serious sources of error will be pick-up on the long
line between the small loop and the receiver. With a coax line there
will be a greater signal pick up on the coax braid than there is in
the loop. They are both located in the same field.

So best to use very low impedance balanced pair line such as 50 ohms
perhaps with a screening braid. A good choke balun or a 1-to-1 wound
transformer would be advisable between the line and receiver input.

Also, depending on frequency, length and impedances, there may be
standing waves on the line which could make a mess of your
calculations.

A change in line length is a good way to check for errors of this
sort.

Fortunately, field strength measurements are seldom needed to great
accuracy. Strength is usually required only to be less than or greater
than some specified value and there is an ample margin for error.

Personally, I think a tuned loop, in the fashion of a magloop, is a
better bet. With its small coupling loop the main loop can be
completely isolated from the line and the line can be ordinary coax
which matches a 50-ohm receiver.

A tuned loop is far more sensitive than the untuned variety. But its
operating frequency range is somwhat restricted.

Field strength measurements are essentially power level measurements
and, ideally, the pick-up loop should be impedance matched to the
receiver. Result : no reflections.
----
Reg.

Reg Edwards wrote:
One of the most serious sources of error will be pick-up on the long
line between the small loop and the receiver. With a coax line there
will be a greater signal pick up on the coax braid than there is in
the loop. They are both located in the same field.

So best to use very low impedance balanced pair line such as 50 ohms
perhaps with a screening braid. A good choke balun or a 1-to-1 wound
transformer would be advisable between the line and receiver input.


Agreed. Because of the inherent balance of the whole loop I have use a
"Voltage Balun", see http://www.vk1od.net/bpl/loop.jpg . I have made
observations of the received signal level when close to aerial telephone
lines carrying ADSL and the pickup level seems the same no matter which
side of the loop is nearest the aerial line.

Also, depending on frequency, length and impedances, there may be
standing waves on the line which could make a mess of your
calculations.

I have made the assumption that the line is adequately terminated in 50
ohms (the attenuator, and there should be now standing waves. Doesn't
that seem reasonable?


A change in line length is a good way to check for errors of this
sort.

Fortunately, field strength measurements are seldom needed to great
accuracy. Strength is usually required only to be less than or greater
than some specified value and there is an ample margin for error.

Personally, I think a tuned loop, in the fashion of a magloop, is a
better bet. With its small coupling loop the main loop can be
completely isolated from the line and the line can be ordinary coax
which matches a 50-ohm receiver.

A tuned loop is far more sensitive than the untuned variety. But its
operating frequency range is somwhat restricted.


Noted.

I have encouraged another ham friend to design an active loop with an AF
good enough to get the system noise floor below -10dBuV at 7MHz. That
is another alternative, and it has issues I know.

I am also considering trying to measure the performance of a portable
short dipole such as a buddipole ( http://www.buddipole.com/ ) for the
purposes of measurement down to ambient noise and a little lower.

Field strength measurements are essentially power level measurements
and, ideally, the pick-up loop should be impedance matched to the
receiver. Result : no reflections.

But if the rx terminates the line, does it matter whether the
"generator" impedance is matched? (I am not trying to bait anyone here,
but Reg, I think I understand the standing wave issue you are raising,
but my reasoning is that if the rx terminates the line sufficiently
well, then standing wave ratio will be small and the error contribution
negligible.)

(I think the lights have gone out on the other side of the big pond.)

Owen

Owen,
I gather you are interested in measurents only in the 40m band which
makes life easier. The photograph of the loop and line makes it more
clear what you are up to.

Yes, there will be no standing waves on the line if the line Zo is
equal to receiver input impedance. (I didn't make myself clear). If a
balun is used it doesn't matter much what Zo is, provided the balun
has the correct ratio. So it is necessary to know what Zo actually is
just as accurately as the input impedance is known.

With your setup it is impossible to match the loop to the line. But if
it WAS possible (eg., as with a magloop) it would NOT be to prevent
standing waves.

Your calculations take the loop/line mismatch loss into account. Its
only a few dB.

Incidentally, have you considered what effects increasing the number
of turns to 2 or 3 would have? They MIGHT possibly be beneficial. It
needs more calculations. As with just increasing the size of the loop
which almost certainly would be beneficial.

You have set yourself a most interesting and useful task. I wish you
well with it.
----
Reg, G4FGQ

On Sun, 19 Jun 2005 01:59:51 GMT, Owen wrote:

Hi Owen,

Between:

I have looked at it and I can't see that I said "power" in relation to
Antenna Factor. Perhaps I am blind. (You didn't confuse the units dB/m
(dB per meter) with dBm (db wrt 1mW) did you?)

and:

I have gotten sidetracked here, my real interest is the completeness /
accuracy of the loop model.

I have observed that your Mathcad design is short of providing
something of a self-check feature. No where in any of your formulas
do you use Units.

Yes, you label them as notes, but this is risky and has been revealed
in your first comments in response to Ed's comment:
Looks pretty decent, until the very end. Antenna Factor (AF) is the ratio of
the field strength voltage to the output VOLTAGE, not power, although you
did get the numbers right.

Folks who follow your math work, will be skimming it, and perhaps a
few will be transcribing it while others will have picked up your MCD
file. This is to say, very few will actually go the whole distance
for a sanity check.

That sanity check is to include the Units within the formulaic Mathcad
expression; that is, after all, one of the boons of using this
package, otherwise any spread sheet would do as well. This inclusion
would enforce a strict compliance with keeping every transformation
accurate, and you would not end up mixing terms which is very simple
to do - and later suffer from. Cecil's work with Photonics suffers
from this problem horribly such that expressions of power end up in
terms so bizarre and thoroughly out of whack that anything could be
proven, except the proof.

73's
Richard Clark, KB7QHC

No you won't, I have posted a later version of the mathcad file to
http://www.vk1od.net/bpl/loop02.mcd .

Yes, got it ok, thanks.

The file you downloaded is an image (.gif) called loop.mcd.gif, and it
looks like your download process dropped the extension, or you hide the
extension on your machine. (Some software thinks that the first dot begins
the file extension, whereas it is the last dot that does so.)

Guessed it was something like that.

Might be interesting to replicate your results with NEC2.

I have modeled the loop in EZNEC and get very similar inductance and
resistances.

Now this is where it gets interesting, and hope I can learn something from
it, as I am sure I have made a mistake someplace. I have not directly
attempted to verify your math, so don't know if you developed it from first
principals or got it from a book. I have a number of references including
Kraus' "Antennas", and also a text by Stutzman and Thiel, etc., so may try
to replicate your methods later.

Using NEC2 I set up a 40 m dipole in free space, and fed it with 1 kW (for
nice large current values in the loop). I placed a square loop, 0.5 m per
side and 40 m distance. with the plane of the loop parallel to the axis of
the dipole, also two sides parallel to the dipole. The dipole uses perfect
conductors, and copper for the loop, with 0.7 mm radius conductors. The
segmentation of the loop is significantly different than the dipole, but
thought it not important because of the large separation of the two
antennas. In the loop I am very close to the minimum segmentation allowed
in NEC at 0.001 wavelengths per segment -- i.e. 11 segments per side. One
segment, near a corner, has a 50 ohm load.

As is easily verified, the field strength from the dipole, at 40 m, is
5.5V/m (RMS). According to NEC the current in the loop varies from segment
to segment, ranging from 0.1 mA (peak), to 0.3 mA (peak). I took the
average (0.191 mA peak), and computed the RMS average current in the loop.
Multiplying by 50 ohms, gives me an output voltage 6.76 mV RMS.

The antenna factor is therefore 58 dB. Wonder if anybody has any idea where
the error lies. I have copied the code below.

Regards,

Frank

NEC Code:

CM Dipole antenna
CE
GW 1 41 0 0 0 20.25 0 0 0.0026706
GW 2 11 10 40 0.25 10.5 40 0.25 0.0007
GW 3 11 10.5 40 0.25 10.5 40 -0.25 0.0007
GW 4 11 10.5 40 -0.25 10 40 -0.25 0.0007
GW 5 11 10 40 -0.25 10 40 0.25 0.0007
GS 0 0 1
GE 0
EX 0 1 21 0 379.63 0.00000
LD 4 5 1 1 50 0
LD 5 2 1 11 5.8001E7
LD 5 3 1 11 5.8001E7
LD 5 4 1 11 5.8001E7
LD 5 5 1 11 5.8001E7
FR 0 12 0 0 7.15 0.0025
RP 0 181 1 1000 -90 90 1 1
EN

Reg Edwards wrote:
Owen,
I gather you are interested in measurents only in the 40m band which
makes life easier. The photograph of the loop and line makes it more
clear what you are up to.

Not necessarily, but the exploration of the loop has been done on lower
HF, and it happens that the BPL system that I have available for
measurement radiates on 7 and 10MHz in low HF.

7MHz is not the only band affected, these guys will use every scratch of
spectrum and power to maximise the speed / reach profile of their service.

....

Your calculations take the loop/line mismatch loss into account. Its
only a few dB.

Incidentally, have you considered what effects increasing the number
of turns to 2 or 3 would have? They MIGHT possibly be beneficial. It
needs more calculations. As with just increasing the size of the loop
which almost certainly would be beneficial.


I did.

It obviously increases the open circuit voltage. It also increases the
loop inductance, and this almost completely offsets the increased open
circuit voltage in terms of power delivered to the receiver input
depending on frequency). Calculation of the wire loss resistance becomes
more complex due to proximity effect, but that doesn't matter too much
because the dominant factor in determining the source Z is the
inductance of the loop, and even if tuned, the resistance is small wrt
the load.

Ofcom had the answer to measurements down to ambient noise level, the
antenna is shown in their recent reports on BPL radiation measurements.
However, it isn't a very portable answer.

As I said in an earlier post, an active loop and a portable short dipole
(such as the Buddipole) are avenues for investigation.

(A tuned loop obviously helps, but with the single frequency /
calibration issues.)

You have set yourself a most interesting and useful task. I wish you
well with it.

Thanks Reg, and I appreciate your help with the task in this discussion
/ review. Wish Amateur Radio well with it, because BPL is the greatest
risk to HF Amateur Radio that we have known. I don't say that from
having read or heard somone else's reports, I have stood on the streets
where BPL is deployed and measured it.

Though my measurement methodology has progressed from "calibrated
S-meter" readings, the calibrated S-meter is a reality check, and when I
last visited the trial site, set the receiver up and waved the 0.5m sq
loop (~-50dBi) about to see if they were still "on air", S-meter
readings of 5uV says they are, and it is seriously high in level.

Ofcom's recent reports are a great read, and it looks like they are
taking a sane approach at this point, differently to the fervour for BPL
expressed by Powell when at the FCC.

I better stop at that, I am getting OT!

Owen

"Owen" wrote in message
...
Ed Price wrote:

Looks pretty decent, until the very end. Antenna Factor (AF) is the ratio
of the field strength voltage to the output VOLTAGE, not power, although
you did get the numbers right.

I have looked at it and I can't see that I said "power" in relation to
Antenna Factor. Perhaps I am blind. (You didn't confuse the units dB/m (dB
per meter) with dBm (db wrt 1mW) did you?)

Owen


Yep, that's exactly what I did. Maybe I was looking at your units too fast
and didn't see that "/" in there.

--
Ed
WB6WSN
El Cajon, CA USA

"Owen" wrote in message
...
Ed Price wrote:

So practically, since the average ham has a receiver with a sensitivity
in the order of a microvolt, then your antenna limits your minimum
discernable signal level to around 65 uV/m. Maybe 100 uV/m to be on the
safe side.

In fact, the technique calls for measuring signals on the rx from the
noise floor to about 20dB above it. The noise floor for receivers today is
typically -135dBm.

No, the technique does not use an S-meter. In a nutshell, it uses Ed
Hare's (W1RFI) technique for calibrating the noise floor of the receiver,
using an external attenuator to keep the rx input below the AGC threshold,
and measuring the audio output with signal and the audio output from rx
internal noise as inputs to a calculation of the input signal power.
Applying external attenuator losses, feedline losses and antenna factor
allows calculation of field strength.

Owen

IS Hare's technique published somewhere on the web?

--
Ed
WB6WSN
El Cajon, CA USA

"Reg Edwards" wrote in message
...
One of the most serious sources of error will be pick-up on the long
line between the small loop and the receiver. With a coax line there
will be a greater signal pick up on the coax braid than there is in
the loop. They are both located in the same field.

So best to use very low impedance balanced pair line such as 50 ohms
perhaps with a screening braid. A good choke balun or a 1-to-1 wound
transformer would be advisable between the line and receiver input.

Also, depending on frequency, length and impedances, there may be
standing waves on the line which could make a mess of your
calculations.

A change in line length is a good way to check for errors of this
sort.


Wouldn't it be better to use a pre-amp at the loop feed? The gain of the
pre-amp could make line pickup a negligible effect, and the pre-amp would
match the coax very well.

--
Ed
WB6WSN
El Cajon, CA USA

"Owen" wrote in message
...
Frank wrote:

In the BPL report at
http://www.ofcom.org.uk/research/tec...line/ascom.pdf
I noticed the system noise floor at about 10 dBuV/m (in 9 kHz). For the
tests they used an active bi-conical antenna. (By my calculations 10
dBuV/m is about 9 uV(2.5 kHz BW) from a 40 m dipole at 7 MHz.) In the
previously

Yes, Fig 8 shows about 10dBuV/m in 9KHz which interpolates to 5dBuV/m in
3KHz, and their measurements used a peak detector. On white noise, the QP
value would probably be 2 to 3dB lower.

I have made a large number of measurements at my home QTH (in a
residential neighbourhoos) using a half wave dipole and assuming an
average gain of -1.2dBi or an AF of -11.6dB/m and I regularly get ambient
noise readings down to around 0 to 3dBuV/m QP in 3KHz or extrapolated to
9KHz BW, 5 to 8dBuV/m QP. Ambient noise is probably lower than indicated
by ITU P372-8!

mentioned report most of the BPL signals -- even at 1 meter from the
source -- is 60 dBuV/m. It seems your system with the loop will be
much less sensitive at about 100 uV/m (+40 dBuV/m).


See my response to Reg re the noise floor for the setup, I make it around
8dbuV/m or 2.5uV/m. I don't pretend it can measure ambient noise, but it
can and has measured BPL interference at 40dBuV/m to 70dBuV/m.


What detector do you think should be used to evaluate the interference
potential of BPL?

I had thought that the QP detector was designed to the "annoyance" effect to
AM or SSB modulation. CISPR has standardized this detector, and it's been
adopted for many legal compliance standards world-wide. Yet the USA &
British military insist on use of a Peak detector. Perhaps a dual level is
needed, with a QP value for comparison of harm to the older analog
modulation techniques, and a Peak value, for comparison of harm to digital
modulation techniques.

--
Ed
WB6WSN
El Cajon, CA USA

Ed Price wrote:
"Owen" wrote in message
...

Ed Price wrote:


So practically, since the average ham has a receiver with a sensitivity
in the order of a microvolt, then your antenna limits your minimum
discernable signal level to around 65 uV/m. Maybe 100 uV/m to be on the
safe side.

In fact, the technique calls for measuring signals on the rx from the
noise floor to about 20dB above it. The noise floor for receivers today is
typically -135dBm.


No, the technique does not use an S-meter. In a nutshell, it uses Ed
Hare's (W1RFI) technique for calibrating the noise floor of the receiver,
using an external attenuator to keep the rx input below the AGC threshold,
and measuring the audio output with signal and the audio output from rx
internal noise as inputs to a calculation of the input signal power.
Applying external attenuator losses, feedline losses and antenna factor
allows calculation of field strength.


Owen


IS Hare's technique published somewhere on the web?


Yes it is, see http://www.arrl.org/~ehare/aria/ARIA_MANUAL_TESTING.pdf .

That paper outlines the principle of using the known rx noise floor as a
baseline for measurements. I have developed a piece of software for
making the associated audio power measurements and automating the
calculation / documentation process.

Thanks for taking the time to review the loop model, it is appreciated.

Owen

PS: I saw your other response and Richard's suggestion that I use the
units capability of Mathcad. Sometimes the unit capability gets in the
way of readability, for instance I think you could not take the log of E
in volts / meter divided by V in volts and get dBV/m, I think you would
need to split E into two variables (say E' and l) and say
AF=20*log(E'/V)/l. Additionally, you can spend more time trying to get
the units to work, so that they don't collapse to fundamental units of
MLT etc, than solving the numerical side of the problem.

"Owen" wrote in message
...
Ed Price wrote:
"Owen" wrote in message
...

Ed Price wrote:


So practically, since the average ham has a receiver with a sensitivity
in the order of a microvolt, then your antenna limits your minimum
discernable signal level to around 65 uV/m. Maybe 100 uV/m to be on the
safe side.

In fact, the technique calls for measuring signals on the rx from the
noise floor to about 20dB above it. The noise floor for receivers today
is typically -135dBm.


No, the technique does not use an S-meter. In a nutshell, it uses Ed
Hare's (W1RFI) technique for calibrating the noise floor of the receiver,
using an external attenuator to keep the rx input below the AGC
threshold, and measuring the audio output with signal and the audio
output from rx internal noise as inputs to a calculation of the input
signal power. Applying external attenuator losses, feedline losses and
antenna factor allows calculation of field strength.


Owen


IS Hare's technique published somewhere on the web?


Yes it is, see http://www.arrl.org/~ehare/aria/ARIA_MANUAL_TESTING.pdf .

That paper outlines the principle of using the known rx noise floor as a
baseline for measurements. I have developed a piece of software for making
the associated audio power measurements and automating the calculation /
documentation process.

Thanks for taking the time to review the loop model, it is appreciated.

Owen


Thanks; looks like I have a lot of reading to do!

--
Ed
WB6WSN
El Cajon, CA USA

Wouldn't it be better to use a pre-amp at the loop feed? The gain of
the
pre-amp could make line pickup a negligible effect, and the pre-amp
would
match the coax very well.

--
Ed
WB6WSN
El Cajon, CA USA

It might be. But the extra complication of powering an amplifier would
bring another load of things to worry about. Simplicity is a
wonderful thing.

By far the best way of improving performanc and reducing possible
measuring errors, is to increase size of loop relative to length of
feedline. Doubling dimensions would make a world of difference.

I suppose he had a good reason for choosing a 1/2-metre square loop.
----
Reg, G4FGQ

Reg Edwards wrote:

I suppose he had a good reason for choosing a 1/2-metre square loop.

Sources suggested variously that the model's assumption of uniform
current distribution was reasonable if the side was from 0.1 to 0.03
wavelengths. I chose the go with the more conservative value at this time.

Owen

"Reg Edwards" wrote in message
...

Wouldn't it be better to use a pre-amp at the loop feed? The gain of
the
pre-amp could make line pickup a negligible effect, and the pre-amp
would
match the coax very well.

--
Ed
WB6WSN
El Cajon, CA USA

It might be. But the extra complication of powering an amplifier would
bring another load of things to worry about. Simplicity is a
wonderful thing.

By far the best way of improving performanc and reducing possible
measuring errors, is to increase size of loop relative to length of
feedline. Doubling dimensions would make a world of difference.

I suppose he had a good reason for choosing a 1/2-metre square loop.
----
Reg, G4FGQ

I have done some more calculations on a square loop fed at the corner. I
agree with your results of input impedance. NEC 2 shows Zin at 0.388 +
j109. The radiation efficiency of such a loop is 2.88%. I have made a very
careful analysis of the currents in the loop when in the presence of a know
E field. The current appears to vary in a sinusoidal manner around the
loop, with very slight discontinuities at the corners. I assume the
variation in current is due to the fact that the induced current is
different on those conductors normal to the dipole axis.

Using the RMS current through the 50 ohm load resistor at the corner, and
more careful calculations, I obtain an antenna factor of 60 dB, or 24 dB
more than your findings. It will be interesting to find why we have such a
large difference.

Regards,

Frank

Ed Price wrote:
....
What detector do you think should be used to evaluate the interference
potential of BPL?

I had thought that the QP detector was designed to the "annoyance" effect to
AM or SSB modulation. CISPR has standardized this detector, and it's been
adopted for many legal compliance standards world-wide. Yet the USA &
British military insist on use of a Peak detector. Perhaps a dual level is
needed, with a QP value for comparison of harm to the older analog
modulation techniques, and a Peak value, for comparison of harm to digital
modulation techniques.


I understand that the CISPR 16-1 QP detector and 9KHz bandwidth are
rooted in a series of subjective listening tests done somewhere around
the 1930s.

(cue storyteller here...)

Firstly, the bandwidth survives, and you are right that it is embedded
in all sorts of EM standards.

I understand that part of the tests I referred to was to discover a
instrument response that fitted well with subjective assessment of the
impact of interference (I presume on an AM broadcast transmission).

I think EMC measurement equipment often contains some of Average, RMS,
QP and Peak detectors.

I have seen several recent reports on BPL radiation that have not used
the QP detector and the reasons have IIRC been that on a scan in xyz
planes over a wide range of frequencies, the EMC receiver is too slow
using the QP detectors.

Ed Hare suggests that the AGC on a receiver acts similarly to the QP
detector, and he is probably right. So the effect being that in an
impulse noise scenario, your receiver will reduce gain roughly in line
with the QP value (rather than say the RMS or the Peak), so it may be a
good measure of gain reduction due to interference. As to interference
with the detection process, the subjective tests to arrive at the QP
detector did not assess impact on digital modulation. It seems to me
that the impact on digital modulation / encoding systems would depend on
the peak value / repetition scenario in concert with the encoding
system's capacity for error detection / correction, but that is just me
thinking aloud.

Measurement bandwidth and extrapolation / interpolation is an issue, and
possibly a bigger one than the detector response. Again there is a
disconnect between 9KHz MBW (below 30MHz) for the standards and the 2KHz
wide receivers in use for SSB. In my opinion, there is no better way to
demonstrate the impact of interference in a 2KHz wide receiver than to
measure it on a 2KHz wide receiver, so I suggest that (for us amateurs)
there may be value in measuring both where possible.

Owen

I have done some more calculations on a square loop fed at the corner. I
agree with your results of input impedance. NEC 2 shows Zin at 0.388 +
j109. The radiation efficiency of such a loop is 2.88%. I have made a
very careful analysis of the currents in the loop when in the presence of
a know E field. The current appears to vary in a sinusoidal manner around
the loop, with very slight discontinuities at the corners. I assume the
variation in current is due to the fact that the induced current is
different on those conductors normal to the dipole axis.

Using the RMS current through the 50 ohm load resistor at the corner, and
more careful calculations, I obtain an antenna factor of 60 dB, or 24 dB
more than your findings. It will be interesting to find why we have such
a large difference.

I see nobody picked up my deliberate mistake! Using Terman's simplified
formulas on pp 813, and 814, I get identical parameters of induced voltage,
and antenna factor. I completely agree with the results of the Mathcad
file, although have not gone through the more intricate equation development
yet.

Using NEC 2 I found out, the hard way, that maximum pick-up takes place off
the ends of the loop. Given these constraints: With a field strength of
5.5 V/m, the voltage across the 50 Ohm resistor is 0.091 V RMS, or an
antenna factor of 35.1 dB. Within 0.4 dB of Owen's results.

Interesting exercise, and I actually learned something.

73,

Frank

Frank wrote:

Using NEC 2 I found out, the hard way, that maximum pick-up takes place off
the ends of the loop. Given these constraints: With a field strength of
5.5 V/m, the voltage across the 50 Ohm resistor is 0.091 V RMS, or an
antenna factor of 35.1 dB. Within 0.4 dB of Owen's results.


That is great Frank, I like it because it agrees of course, but if its a
valid model, that is even better.

My model was much simpler in just trying to establish the Z of the loop,
I didn't do what you have done and irradiated the loop, well done. My
only residual concern is whether your receive loop was far enough from
the exciter to be truly operating under free space conditions. I suppose
if you double the distance and get nearly identical results, that would
be sufficient confirmation.

Two questions:
1. can I have a look at your model;
2. would you permit me to publish it (with attribution) on a web page
that I am drafting on the antenna design for BPL interference and other
interference measurement purposes.

BTW, although the gain is really low, I have also used it effectively
for getting bearings on interference from afar (though, DFing HF signals
is problematic, especially at high path angles).

Interesting exercise, and I actually learned something.


Life is fun, isn't it.

There is more to ham radio that "logging-in" to content-free nets,
talking about what is on the menu for dinner. Indeed, I suspect that
more "real" ham radio takes place off-air than on-air.

Owen

Frank wrote:
....
Using NEC 2 I found out, the hard way, that maximum pick-up takes place off
the ends of the loop. Given these constraints: With a field strength of
5.5 V/m, the voltage across the 50 Ohm resistor is 0.091 V RMS, or an
antenna factor of 35.1 dB. Within 0.4 dB of Owen's results.


I have modified my EZNEC model to irrradiate the loop from an exciter
dipole at some distance. I have tried it with the exciter at 100m and
200m distance, I get AF=35.69 and 35.72 dB respectively (against 36.1
from the Mathcad model). I note that if you report on the so called near
field values too close to the loop, the results are effected by the loop.

I note that the Matchcad model estimates the inductance slightly higher
than EzNEC, and others have suggested to me that the Lw term in the
Matchcad model is a double up, that the Ll term fully accounts for the
inductance of the loop. Removal of the Lw term results in an AF of
35.8dB, which is closer to the EzNEC prediction.

Has anyone views of whether Lw should not be included?

The currents in my model vary slightly, here is the current report:

Frequency = 7.1 MHz

Wire No. 1:
Segment Conn Magnitude (A.) Phase (Deg.)
1 W4E2 4.0E-6 -137.0
2 4.0E-6 -137.7
3 4.0E-6 -138.1
4 4.0E-6 -138.4
5 4.0E-6 -138.5
6 4.0E-6 -138.4
7 4.0E-6 -138.1
8 4.0E-6 -137.5
9 W2E1 4.0E-6 -136.8

Wire No. 2:
Segment Conn Magnitude (A.) Phase (Deg.)
1 W1E2 4.1E-6 -136.0
2 4.1E-6 -135.1
3 4.1E-6 -134.2
4 4.1E-6 -133.4
5 4.2E-6 -132.5
6 4.2E-6 -131.7
7 4.2E-6 -130.8
8 4.3E-6 -130.0
9 W3E1 4.3E-6 -129.2

Wire No. 3:
Segment Conn Magnitude (A.) Phase (Deg.)
1 W2E2 4.3E-6 -128.4
2 4.3E-6 -127.8
3 4.4E-6 -127.4
4 4.4E-6 -127.2
5 4.4E-6 -127.1
6 4.4E-6 -127.2
7 4.4E-6 -127.4
8 4.3E-6 -127.9
9 W4E1 4.3E-6 -128.5

Wire No. 4:
Segment Conn Magnitude (A.) Phase (Deg.)
1 W3E2 4.3E-6 -129.3
2 4.3E-6 -130.1
3 4.2E-6 -130.9
4 4.2E-6 -131.8
5 4.2E-6 -132.6
6 4.1E-6 -133.5
7 4.1E-6 -134.3
8 4.1E-6 -135.2
9 W1E1 4.1E-6 -136.1



Owen

I have written a deck for NEC2 which uses and incident plane wave as the
excitation scenario.

The deck:

CMSmall square untuned loop
CMEXTENDED THIN WIRE KERNEL USED
CM1. FREE SPACE
CE2. Plane wave excitation
GW 1 9 -0.25 0 1 -0.25 0 1.5 .0007
GW 2 9 -0.25 0 1.5 +0.25 0 1.5 .0007
GW 3 9 +0.25 0 1.5 +0.25 0 1 .0007
GW 4 9 +0.25 0 1 -0.25 0 1 .0007
GE 1
EK
FR 0 1 0 0 7.1
EX 1 1 1 0 90 0 0 0 0 0
LD 4 1 1 1 50 0
GN -1
XQ
EN

I get an antenna factor of 35.8dB/m by multiplying the current in wire
1, seg 1 (the location of the 50 ohm load) by 50 ohms, and taking
-20*log of that voltage.

Owen

PS: This won't work if you are working with one of the original column
formatted Fortran inputs, but it does work with nec2++ (
http://www.si-list.org/NEC_Archives/swindex.html )

It is futile to accept Mathcad or any other program as being the Bible
unless one is sure that the underlying reasoning, by the user, is
correct.

Validity of the results are dependent solely on the user, plus the
possible uncertainty of program bugs.
----
Reg, G4FGQ

Reg Edwards wrote:
It is futile to accept Mathcad or any other program as being the Bible
unless one is sure that the underlying reasoning, by the user, is
correct.

Validity of the results are dependent solely on the user, plus the
possible uncertainty of program bugs.

Well, perhaps the validity of the model, which leads me to...

Reg, I mentioned in an earlier post that I was concerned about the
estimate of the loop inductance. It was suggested to me that the Lw term
that I included because it was in a prof's lecture notes was a double
up, and that the La term sufficiently estimates the inducance of the loop.

Searching the net suggests that there is some degree of black magic or
black art in estimating inductance of such things, and that Grover and
Terman had a view on it, but the formula I used seems to be commonly
used by online calculators and perhaps a more modern view.

Can you (or others) throw any light on the most appropriate formula for
estimation of the inductance of a square loop of about the size in the
model (0.5m sides, ~2mm dia copper wire).

I do recall that you said what I have done is too complicated and it can
be done with a hand calculator... but I would like to find an expression
for the inductance of the loop that will be accepted generally as the
best estimate withing the other constraints of the model and construction.

All constructive help appreciated.

Owen

Owen wrote:
Reg Edwards wrote:

It is futile to accept Mathcad or any other program as being the Bible
unless one is sure that the underlying reasoning, by the user, is
correct.

Validity of the results are dependent solely on the user, plus the
possible uncertainty of program bugs.


Well, perhaps the validity of the model, which leads me to...

Reg, I mentioned in an earlier post that I was concerned about the
estimate of the loop inductance. It was suggested to me that the Lw term
that I included because it was in a prof's lecture notes was a double
up, and that the La term sufficiently estimates the inducance of the loop.

Searching the net suggests that there is some degree of black magic or
black art in estimating inductance of such things, and that Grover and
Terman had a view on it, but the formula I used seems to be commonly
used by online calculators and perhaps a more modern view.

Can you (or others) throw any light on the most appropriate formula for
estimation of the inductance of a square loop of about the size in the
model (0.5m sides, ~2mm dia copper wire).

I do recall that you said what I have done is too complicated and it can
be done with a hand calculator... but I would like to find an expression
for the inductance of the loop that will be accepted generally as the
best estimate withing the other constraints of the model and construction.

All constructive help appreciated.

Owen

Hi, Owen -

I don't know if the following posts will be of any help at all.

Subjects:
Wheeler's 1982 formulas verified
Wheeler's 1982 formulas verified - Wheeler.gif
Another solenoid inductance calculation - Solenoid.zip

group:
alt.binaries.schematics.electronic

Good luck.

John

"Owen" wrote
Reg, I mentioned in an earlier post that I was concerned about the
estimate of the loop inductance.

Owen,

The HF inductance of a square loop is -

L = 0.8 * H * ( Ln( 4 * H / D ) - 1.467 ) microhenries,

where H is length of one side,

and D is diameter of circular conductor,

both dimensions are in metres.

There are half a dozen other formulas which at first appear to be
different from the above but can be mathematically transformed to be
identical. And then there are imperial and metric units.

I got it out of one of one of my old notebooks. It's what I use in my
programs. I think I stole it from Terman. And Terman stole it from
Grover. So it's sure to be accurate enough for anything you are ever
likely to use it for.

It is obviously an approximation because when a large conductor
diameter is comparable with a very short length of side, the
inductance has a negative value.

As a sanity check, compare it with whatever formula you have used up
to now.
----
Reg, G4FGQ

Reg Edwards wrote:
"Owen" wrote

Reg, I mentioned in an earlier post that I was concerned about the
estimate of the loop inductance.


Owen,

The HF inductance of a square loop is -

L = 0.8 * H * ( Ln( 4 * H / D ) - 1.467 ) microhenries,

where H is length of one side,

and D is diameter of circular conductor,

both dimensions are in metres.


Though it looks a little different, that formula will always produce
exactly the same results as the one that I have used.

There are half a dozen other formulas which at first appear to be
different from the above but can be mathematically transformed to be
identical. And then there are imperial and metric units.

I got it out of one of one of my old notebooks. It's what I use in my
programs. I think I stole it from Terman. And Terman stole it from
Grover. So it's sure to be accurate enough for anything you are ever
likely to use it for.

I looked in Terman, but didn't find it, and still can't. I might be blind!

I got it from http://emcsun.ece.umr.edu/new-induct/ but it obviously
shares the same root as yours. They attribute it to Grover.


It is obviously an approximation because when a large conductor
diameter is comparable with a very short length of side, the
inductance has a negative value.

Understood.


As a sanity check, compare it with whatever formula you have used up
to now.

See above. I think you are telling me I should be confident I am sane.
If it was just that easy!

Now I just have to find someone with access to an OATS to measure one of
these things.

Thanks Reg...

Owen