On 10/21/2015 6:12 AM, amdx wrote:
On 10/20/2015 10:35 PM, rickman wrote:
On 10/20/2015 9:21 PM, amdx wrote:
On 10/20/2015 1:56 PM, rickman wrote:
On 10/20/2015 10:44 AM, amdx wrote:
On 10/19/2015 10:53 PM, rickman wrote:
On 10/19/2015 3:50 PM, bilou wrote:
"rickman" wrote in message
...
On 10/19/2015 3:34 AM, Brian Howie wrote:
How does the coil affect the tuning range of the cap? A cap is
limited by
the ratio of the minimum to maximum capacitance. The ratio of
frequency
is limited to the same ratio.
In a multiturn loop you get huge capacitance between turns.
For a given variable capacitor it appears in parallel.
The Q of that big coil might be higher but as you need to add
fixed capacitors to the variable one to get useful tuning range
you loose almost what you gain.

I sort of lost the thought here. If you up the inductance of the
loop,
it lowers the required tuning capacitance, so why would fixed
capacitors
be needed? Are you saying the parasitic capacitance of the loop is
enough to significantly reduce the tuning range of the variable cap?
Maybe, but there are construction methods that minimize the parasitic
capacitance of multi-turn loops. Wide spacing is important. I've
seen
spiral loops wound on wooden frames that look like God's Eyes, very
attractive.


I saw descriptions using a 128 pairs telephone cable and spending
several days to wire it as a 256 turns loop.
A bad idea IMHO.

I'm not sure what problem you would be trying to solve by using a 256
turn loop. There are middle grounds...


Often a 60kHz WWVB time receiver.

So why would that be a "bad idea"?


Ahh, you ask "what problem you would be trying to solve"
I should clarify, a resonant antenna for 60kHz, and that requires a
large inductance. Or at least that is one approach.

But the context was that a 256 turn loop was a bad thing. I'm trying to
understand what that was about. I don't need to know when it is a good
idea... well, I guess even that is interesting. But I think the way a
256 turn loop would be made for a WWVB receiver is around a piece of
ferrite. But who knows, maybe a large loop of telephone cable would
work well too.

It obviously works. It is not ideal because it would have a lot of
interwinding capacitance. Also the interwinding capacitance is not a
quality capacitance thus the Q is lowered.
It could be built with space between wire and layers, and 256 solder
connections is not a great idea when trying to insure high Q.
As far as "bad idea", all it has to do is receive enough signal
to keep the clock accurate, more than that is interesting, but useless.

I haven't built a high Q antenna yet, but I am pretty sure people
greatly exaggerate the significance of solder connections in the Q
factor. Q is related to the losses. I am sure the solder connections
will not significantly impact the dissipative resistance of the wire
unless the turns are around a pencil.

As to "It obviously works", that depends on many other factors. Sure,
no doubt it will work a mile from the transmitter. What about along the
US east coast in a metal building with many appliances around? There is
working, and there is working well. The inter-winding capacitance is
not a factor as long as the station can be tuned.

--

Rick

On 10/21/2015 4:42 AM, Brian Howie wrote:
In message , rickman writes

The capacitance of the loop to the screen meant that at the minimum
variable C setting ,I couldn't get the maximum frequency of about
500KHz I wanted, so I had to take turns off. I now need more parallel C
to tune the look down to 136KHz.

Wow, that loop must have a *lot* of capacitance. Is there a way to
space the conductors away from the copper tubing in the run?

Not easy

I'm curious why you would use copper pipe for the shield. Because it
provides both shield and support? I guess there are a million ways to
build a shielded loop. I like the idea of using coax, but I don't
know if that also has serious limitations from the capacitance between
loop conductor and shield.


It seemed a good idea at the time. The original design used plastic pipe
covered with tin-foil ,but I wanted something that would survive a
Scottish winter outdoors.

PVC 4-7 Loop Antenna Al Burzynski KA5JGV ( it's on the NDB yahoo group)

it used 12 turns. I think the use of plastic pipe and external tinfoil
reduces the C.

My loop does work quite well, and has survived outdoors but I think it
could be improved

Yes, well, they can *all* be improved. I find it interesting to make
stuff like this from discarded materials like rain gutter. I'm trying
to understand why one poster in a yahoo group says thin stock is no good
for transmitting loop antennas. The skin effect limits the signal to
the outer few mils of copper or aluminum. I think thin stock will do
just fine if the circumference is large, overcoming the limitations of
the skin effect. At least you are using all the material rather than
just the outer few mils.

--

Rick

On 10/21/2015 6:15 AM, amdx wrote:
On 10/21/2015 2:06 AM, rickman wrote:
On 10/21/2015 2:18 AM, Brian Howie wrote:
In message , rickman
writes
On 10/19/2015 3:34 AM, Brian Howie wrote:
In message , bilou
writes

"Brian Howie" wrote in message
...

I've a 5 foot Octagonal loop for MF. The shield is copper water
pipe,
with
a gap , 7 turns inside plus a coupling winding. It does a good job
eliminating local noise (mostly ASDL hash from the phone lines)
compared
with a vertical. However the capacitance between the shield and
turns
seems to load it quite a bit meaning I can't get the tuning range
I'd
like.

Brian GM4DIJ
--
Brian Howie
Hi
My own experience is that ,at least for receive, multi turn loops are
useless.
Instead you can use a single turn one with a good coil in serial.
The tuning range for a given variable capacitor is much greater
especially if ,at low frequency, the coil is using ferrite .
Switching the coil can increase the tuning range easily.
The coil, with a secondary winding,is also very useful to
adjust the coupling to the receiver.

I'd have thought I'd get a better signal from more turns, but maybe
better coupling and a higher Q from your suggestion would do the
same.

I can't imagine why more turns won't help a receiving loop. I guess
it depends on what is limiting reception. Adding a coil may improve
the Q or it make make it worse depending on the Q of the coil. More
turns won't help the Q of a receiving loop, other than reducing the
significance of the resistance of connections and other components.
More turns *will* increase the signal strength.

How does the coil affect the tuning range of the cap? A cap is
limited by the ratio of the minimum to maximum capacitance. The ratio
of frequency is limited to the same ratio.


The capacitance of the loop to the screen meant that at the minimum
variable C setting ,I couldn't get the maximum frequency of about
500KHz I wanted, so I had to take turns off. I now need more parallel C
to tune the look down to 136KHz.

Wow, that loop must have a *lot* of capacitance. Is there a way to
space the conductors away from the copper tubing in the run?

I'm curious why you would use copper pipe for the shield. Because it
provides both shield and support? I guess there are a million ways to
build a shielded loop. I like the idea of using coax, but I don't know
if that also has serious limitations from the capacitance between loop
conductor and shield.

30pf per ft is a general number for capacitance of coax, but you know
it varies with type. I have some coax for automobile radio antennas
(AM/FM) that has 8pf per foot.

Doesn't the capacitance vary mostly with diameter? The RG-6 I have is
16 pf/ft about a quarter inch diameter.

--

Rick

On 10/21/2015 2:36 PM, rickman wrote:
On 10/21/2015 6:12 AM, amdx wrote:
On 10/20/2015 10:35 PM, rickman wrote:
On 10/20/2015 9:21 PM, amdx wrote:
On 10/20/2015 1:56 PM, rickman wrote:
On 10/20/2015 10:44 AM, amdx wrote:
On 10/19/2015 10:53 PM, rickman wrote:
On 10/19/2015 3:50 PM, bilou wrote:
"rickman" wrote in message
...
On 10/19/2015 3:34 AM, Brian Howie wrote:
How does the coil affect the tuning range of the cap? A cap is
limited by
the ratio of the minimum to maximum capacitance. The ratio of
frequency
is limited to the same ratio.
In a multiturn loop you get huge capacitance between turns.
For a given variable capacitor it appears in parallel.
The Q of that big coil might be higher but as you need to add
fixed capacitors to the variable one to get useful tuning range
you loose almost what you gain.

I sort of lost the thought here. If you up the inductance of the
loop,
it lowers the required tuning capacitance, so why would fixed
capacitors
be needed? Are you saying the parasitic capacitance of the loop is
enough to significantly reduce the tuning range of the variable cap?
Maybe, but there are construction methods that minimize the
parasitic
capacitance of multi-turn loops. Wide spacing is important. I've
seen
spiral loops wound on wooden frames that look like God's Eyes, very
attractive.


I saw descriptions using a 128 pairs telephone cable and spending
several days to wire it as a 256 turns loop.
A bad idea IMHO.

I'm not sure what problem you would be trying to solve by using a
256
turn loop. There are middle grounds...


Often a 60kHz WWVB time receiver.

So why would that be a "bad idea"?


Ahh, you ask "what problem you would be trying to solve"
I should clarify, a resonant antenna for 60kHz, and that requires a
large inductance. Or at least that is one approach.

But the context was that a 256 turn loop was a bad thing. I'm trying to
understand what that was about. I don't need to know when it is a good
idea... well, I guess even that is interesting. But I think the way a
256 turn loop would be made for a WWVB receiver is around a piece of
ferrite. But who knows, maybe a large loop of telephone cable would
work well too.

It obviously works. It is not ideal because it would have a lot of
interwinding capacitance. Also the interwinding capacitance is not a
quality capacitance thus the Q is lowered.
It could be built with space between wire and layers, and 256 solder
connections is not a great idea when trying to insure high Q.
As far as "bad idea", all it has to do is receive enough signal
to keep the clock accurate, more than that is interesting, but useless.

I haven't built a high Q antenna yet, but I am pretty sure people
greatly exaggerate the significance of solder connections in the Q
factor. Q is related to the losses. I am sure the solder connections
will not significantly impact the dissipative resistance of the wire
unless the turns are around a pencil.



On a large loop antenna, it is probably difficult to get an extremely

high Q. So, the solder connections will have less of an effect than if
it was higher.
I made a loop with 1/4" copper pipe, about 2'x 2' with a vacuum variable.
I measured it at about Q=800.

Using a 240uh and assuming 1000kHz, That's about 1.88 ohms of loss,
split between dissipation in materials,
wire losses, connection losses and capacitor losses.
If you had 0.12 ohms additional solder connection losses, Q would drop
to 753 from 800.




As to "It obviously works", that depends on many other factors. Sure,
no doubt it will work a mile from the transmitter. What about along the
US east coast in a metal building with many appliances around? There is
working, and there is working well. The inter-winding capacitance is
not a factor as long as the station can be tuned.

It is my pet theory, that interwinding capacitance will lower Q.
It causes displacement current which causes more current flow between
turns, also, the interwinding capacitance causes capacitive proximity
effects. (vs magnetic proximity effect)
I suppose a neat experiment would be to find two materials with equal
losses but one having much higher permittivity. Then test Q with one
material placed between turns, pull that out and install the higher
permittivity material and retest Q.

But, I could be all wrong on the subject.

Check out this guys site, has some nice loops.
http://makearadio.com/loops/

Mikek

On 10/21/2015 2:47 PM, rickman wrote:
On 10/21/2015 6:15 AM, amdx wrote:
On 10/21/2015 2:06 AM, rickman wrote:
On 10/21/2015 2:18 AM, Brian Howie wrote:
In message , rickman
writes
On 10/19/2015 3:34 AM, Brian Howie wrote:
In message , bilou
writes

"Brian Howie" wrote in message
...

I've a 5 foot Octagonal loop for MF. The shield is copper water
pipe,
with
a gap , 7 turns inside plus a coupling winding. It does a good job
eliminating local noise (mostly ASDL hash from the phone lines)
compared
with a vertical. However the capacitance between the shield and
turns
seems to load it quite a bit meaning I can't get the tuning range
I'd
like.

Brian GM4DIJ
--
Brian Howie
Hi
My own experience is that ,at least for receive, multi turn loops
are
useless.
Instead you can use a single turn one with a good coil in serial.
The tuning range for a given variable capacitor is much greater
especially if ,at low frequency, the coil is using ferrite .
Switching the coil can increase the tuning range easily.
The coil, with a secondary winding,is also very useful to
adjust the coupling to the receiver.

I'd have thought I'd get a better signal from more turns, but maybe
better coupling and a higher Q from your suggestion would do the
same.

I can't imagine why more turns won't help a receiving loop. I guess
it depends on what is limiting reception. Adding a coil may improve
the Q or it make make it worse depending on the Q of the coil. More
turns won't help the Q of a receiving loop, other than reducing the
significance of the resistance of connections and other components.
More turns *will* increase the signal strength.

How does the coil affect the tuning range of the cap? A cap is
limited by the ratio of the minimum to maximum capacitance. The ratio
of frequency is limited to the same ratio.


The capacitance of the loop to the screen meant that at the minimum
variable C setting ,I couldn't get the maximum frequency of about
500KHz I wanted, so I had to take turns off. I now need more parallel C
to tune the look down to 136KHz.

Wow, that loop must have a *lot* of capacitance. Is there a way to
space the conductors away from the copper tubing in the run?

I'm curious why you would use copper pipe for the shield. Because it
provides both shield and support? I guess there are a million ways to
build a shielded loop. I like the idea of using coax, but I don't know
if that also has serious limitations from the capacitance between loop
conductor and shield.

30pf per ft is a general number for capacitance of coax, but you know
it varies with type. I have some coax for automobile radio antennas
(AM/FM) that has 8pf per foot.

Doesn't the capacitance vary mostly with diameter?

Distance between conductors and material between.


The RG-6 I have is 16 pf/ft about a quarter inch diameter.

Foamed PE! But in general, it looks like my 30pf is a little high.

Mikek

On 10/21/2015 8:41 PM, amdx wrote:
On 10/21/2015 2:36 PM, rickman wrote:
On 10/21/2015 6:12 AM, amdx wrote:
On 10/20/2015 10:35 PM, rickman wrote:
On 10/20/2015 9:21 PM, amdx wrote:
On 10/20/2015 1:56 PM, rickman wrote:
On 10/20/2015 10:44 AM, amdx wrote:
On 10/19/2015 10:53 PM, rickman wrote:
On 10/19/2015 3:50 PM, bilou wrote:
"rickman" wrote in message
...
On 10/19/2015 3:34 AM, Brian Howie wrote:
How does the coil affect the tuning range of the cap? A cap is
limited by
the ratio of the minimum to maximum capacitance. The ratio of
frequency
is limited to the same ratio.
In a multiturn loop you get huge capacitance between turns.
For a given variable capacitor it appears in parallel.
The Q of that big coil might be higher but as you need to add
fixed capacitors to the variable one to get useful tuning range
you loose almost what you gain.

I sort of lost the thought here. If you up the inductance of the
loop,
it lowers the required tuning capacitance, so why would fixed
capacitors
be needed? Are you saying the parasitic capacitance of the loop is
enough to significantly reduce the tuning range of the variable
cap?
Maybe, but there are construction methods that minimize the
parasitic
capacitance of multi-turn loops. Wide spacing is important. I've
seen
spiral loops wound on wooden frames that look like God's Eyes, very
attractive.


I saw descriptions using a 128 pairs telephone cable and spending
several days to wire it as a 256 turns loop.
A bad idea IMHO.

I'm not sure what problem you would be trying to solve by using a
256
turn loop. There are middle grounds...


Often a 60kHz WWVB time receiver.

So why would that be a "bad idea"?


Ahh, you ask "what problem you would be trying to solve"
I should clarify, a resonant antenna for 60kHz, and that requires a
large inductance. Or at least that is one approach.

But the context was that a 256 turn loop was a bad thing. I'm
trying to
understand what that was about. I don't need to know when it is a good
idea... well, I guess even that is interesting. But I think the way a
256 turn loop would be made for a WWVB receiver is around a piece of
ferrite. But who knows, maybe a large loop of telephone cable would
work well too.

It obviously works. It is not ideal because it would have a lot of
interwinding capacitance. Also the interwinding capacitance is not a
quality capacitance thus the Q is lowered.
It could be built with space between wire and layers, and 256 solder
connections is not a great idea when trying to insure high Q.
As far as "bad idea", all it has to do is receive enough signal
to keep the clock accurate, more than that is interesting, but useless.

I haven't built a high Q antenna yet, but I am pretty sure people
greatly exaggerate the significance of solder connections in the Q
factor. Q is related to the losses. I am sure the solder connections
will not significantly impact the dissipative resistance of the wire
unless the turns are around a pencil.



On a large loop antenna, it is probably difficult to get an extremely

high Q. So, the solder connections will have less of an effect than if
it was higher.
I made a loop with 1/4" copper pipe, about 2'x 2' with a vacuum variable.
I measured it at about Q=800.

Using a 240uh and assuming 1000kHz, That's about 1.88 ohms of loss,
split between dissipation in materials,
wire losses, connection losses and capacitor losses.
If you had 0.12 ohms additional solder connection losses, Q would drop
to 753 from 800.

Why would you assume 120 mOhms of resistance from soldered connections?
Have you measured this somewhere?


As to "It obviously works", that depends on many other factors. Sure,
no doubt it will work a mile from the transmitter. What about along the
US east coast in a metal building with many appliances around? There is
working, and there is working well. The inter-winding capacitance is
not a factor as long as the station can be tuned.

It is my pet theory, that interwinding capacitance will lower Q.
It causes displacement current which causes more current flow between
turns, also, the interwinding capacitance causes capacitive proximity
effects. (vs magnetic proximity effect)
I suppose a neat experiment would be to find two materials with equal
losses but one having much higher permittivity. Then test Q with one
material placed between turns, pull that out and install the higher
permittivity material and retest Q.

But, I could be all wrong on the subject.

Check out this guys site, has some nice loops.
http://makearadio.com/loops/

Yeah, not bad.

--

Rick

On 10/21/2015 9:09 PM, rickman wrote:
On 10/21/2015 8:41 PM, amdx wrote:
On 10/21/2015 2:36 PM, rickman wrote:
On 10/21/2015 6:12 AM, amdx wrote:
On 10/20/2015 10:35 PM, rickman wrote:
On 10/20/2015 9:21 PM, amdx wrote:
On 10/20/2015 1:56 PM, rickman wrote:
On 10/20/2015 10:44 AM, amdx wrote:
On 10/19/2015 10:53 PM, rickman wrote:
On 10/19/2015 3:50 PM, bilou wrote:
"rickman" wrote in message
...
On 10/19/2015 3:34 AM, Brian Howie wrote:
How does the coil affect the tuning range of the cap? A cap is
limited by
the ratio of the minimum to maximum capacitance. The ratio of
frequency
is limited to the same ratio.
In a multiturn loop you get huge capacitance between turns.
For a given variable capacitor it appears in parallel.
The Q of that big coil might be higher but as you need to add
fixed capacitors to the variable one to get useful tuning range
you loose almost what you gain.

I sort of lost the thought here. If you up the inductance of the
loop,
it lowers the required tuning capacitance, so why would fixed
capacitors
be needed? Are you saying the parasitic capacitance of the
loop is
enough to significantly reduce the tuning range of the variable
cap?
Maybe, but there are construction methods that minimize the
parasitic
capacitance of multi-turn loops. Wide spacing is important. I've
seen
spiral loops wound on wooden frames that look like God's Eyes,
very
attractive.


I saw descriptions using a 128 pairs telephone cable and spending
several days to wire it as a 256 turns loop.
A bad idea IMHO.

I'm not sure what problem you would be trying to solve by using a
256
turn loop. There are middle grounds...


Often a 60kHz WWVB time receiver.

So why would that be a "bad idea"?


Ahh, you ask "what problem you would be trying to solve"
I should clarify, a resonant antenna for 60kHz, and that requires a
large inductance. Or at least that is one approach.

But the context was that a 256 turn loop was a bad thing. I'm
trying to
understand what that was about. I don't need to know when it is a
good
idea... well, I guess even that is interesting. But I think the way a
256 turn loop would be made for a WWVB receiver is around a piece of
ferrite. But who knows, maybe a large loop of telephone cable would
work well too.

It obviously works. It is not ideal because it would have a lot of
interwinding capacitance. Also the interwinding capacitance is not a
quality capacitance thus the Q is lowered.
It could be built with space between wire and layers, and 256 solder
connections is not a great idea when trying to insure high Q.
As far as "bad idea", all it has to do is receive enough signal
to keep the clock accurate, more than that is interesting, but useless.

I haven't built a high Q antenna yet, but I am pretty sure people
greatly exaggerate the significance of solder connections in the Q
factor. Q is related to the losses. I am sure the solder connections
will not significantly impact the dissipative resistance of the wire
unless the turns are around a pencil.



On a large loop antenna, it is probably difficult to get an extremely

high Q. So, the solder connections will have less of an effect than if
it was higher.
I made a loop with 1/4" copper pipe, about 2'x 2' with a vacuum variable.
I measured it at about Q=800.

Using a 240uh and assuming 1000kHz, That's about 1.88 ohms of loss,
split between dissipation in materials,
wire losses, connection losses and capacitor losses.
If you had 0.12 ohms additional solder connection losses, Q would drop
to 753 from 800.

Why would you assume 120 mOhms of resistance from soldered connections?
Have you measured this somewhere?


I have absolutely no idea about the resistance of a solder connection.
I used 0.12 because it made 1.88 = 2.


As to "It obviously works", that depends on many other factors. Sure,
no doubt it will work a mile from the transmitter. What about along the
US east coast in a metal building with many appliances around? There is
working, and there is working well. The inter-winding capacitance is
not a factor as long as the station can be tuned.

It is my pet theory, that interwinding capacitance will lower Q.
It causes displacement current which causes more current flow between
turns, also, the interwinding capacitance causes capacitive proximity
effects. (vs magnetic proximity effect)
I suppose a neat experiment would be to find two materials with equal
losses but one having much higher permittivity. Then test Q with one
material placed between turns, pull that out and install the higher
permittivity material and retest Q.

But, I could be all wrong on the subject.

Check out this guys site, has some nice loops.
http://makearadio.com/loops/

Yeah, not bad.

On Wed, 21 Oct 2015 05:15:18 -0500, amdx wrote:


30pf per ft is a general number for capacitance of coax, but you know
it varies with type. I have some coax for automobile radio antennas
(AM/FM) that has 8pf per foot.
Mikek


Well its not real coax , but the diameter is about 5ft ( well its
octagonal ) say 15ft circumferance , There's 7 turns in close
proximity to the tube or each other, so the capacitance could be a few
hundred pf . I suppose could try measuring it. Self resonance with
12 turns was about 400KHz

http://www.angelfire.com/mb/amandx/loop.html


so effective minimumum capacitance works out around 100pf

Brian


---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

On 10/22/2015 9:43 AM, Brian Howie wrote:
On Wed, 21 Oct 2015 05:15:18 -0500, amdx wrote:


30pf per ft is a general number for capacitance of coax, but you know
it varies with type. I have some coax for automobile radio antennas
(AM/FM) that has 8pf per foot.
Mikek


Well its not real coax , but the diameter is about 5ft ( well its
octagonal ) say 15ft circumferance , There's 7 turns in close
proximity to the tube or each other, so the capacitance could be a few
hundred pf . I suppose could try measuring it. Self resonance with
12 turns was about 400KHz

http://www.angelfire.com/mb/amandx/loop.html


so effective minimumum capacitance works out around 100pf

Brian


I'm not sure if you're discussing 100pf of interwinding capacitance or
capacitance between the shield and the winding.

Mikek

On 10/15/2015 5:42 AM, J.B. Wood wrote:
On 10/14/2015 02:34 PM, rickman wrote:
I just read the wikipedia article on small loop antennas and it seems I
was laboring under a misapprehension. I thought receiving loops were
"magnetic" because they were shielded (this is often stated in various
web pages about constructing such loops). But the wikipedia article on
small loop antennas says the nature of a small loop is to not be very
sensitive to the E field in near field.

So if the shield has little to do with rejecting near field electrical
noise, what does the shield do? A lot of antenna designs make a big
deal of the shield. So I assume it must be a useful addition to the
small loop antenna for some purpose.

Hello, and that seems to be ham radio jargon. Hams seem to think the
adjectives "magnetic" and "electric" are needed when referring to loop
and dipole antennas, respectively. Textbooks on electromagnetics and
antennas don't use those terms except in the case when discussing
theoretically small radiators, i.e. "magnetic dipoles" and "electric
dipoles".

My hypothesis on the ham terminology is that a loop is viewed as an
inductor. That's OK for close-in (non-radiative) mutual coupling to
some source but when you're several wavelengths away (in the far field)
then the loop (or dipole antenna for that matter) responds to the
electromagnetic field (the electric and magnetic far fields can't be
considered separately). The fact that an axis of either antenna lines
up with the electric or magnetic field vector in the far field is moot.
Does this mean that the loop doesn't have inductance? Of course not
and it plays a role in establishing the feedpoint impedance of the loop
at the operating frequency. Now if folks would just stop using that
word "literally" so damn much...

Sincerely, and 73s from N4GG0

I agree. If the jargon is either magnetic or electric, how do we define
a folded dipole antenna? It is a loop. Is it electric or magnetic?